And although the world population grew by 18.6% between 2000 and 2015, the population of these areas exceeded that growth, increasing by 34.1% during the same period. That means that between 58 and 86 million more people were exposed to flooding in those places over the course of 15 years.
“It is not particularly surprising that flooding is increasing,” he says. Beth tellman, co-founder of flood mapping startup From the cloud to the street and the lead author of the study. “But what struck me was that people were moving to places where we had seen flooding in the past.”
The researchers observed more than 3,000 events in the Dartmouth Flood Observatory database, which records floods reported in media coverage. They matched events that had location data with satellite images from FASHION, an instrument mounted on two NASA satellites that have captured daily images of Earth since 2000.
The researchers used an algorithm to map where the floods had occurred by classifying which pixels were covered in water and which were not. They then added population data to see how trends in flooded areas changed over time.
Low- and middle-income countries had the fastest population growth in flood-prone areas in the past two decades, with the highest growth rates in sub-Saharan Africa and South Asia.
Socio-economic factors could explain part of the move, Tellman says. Vulnerable groups may have no choice but to settle in floodplains, where land may be cheaper and more available.
Using satellite imagery, the researchers were able to describe actual flood impacts more accurately than traditional models. Models can capture some types of floods, such as those that occur around rivers and coasts. But for others caused by heavy rain or random events, such as dam ruptures or a storm surge that lines up with high tide, satellite images provide a clearer picture.
The 913 mapped floods are still just a fraction of the tens of thousands that occur globally each year. “It’s just the tip of the iceberg,” Tellman says.
MODIS takes images with a resolution of 250 meters, about the length of two football fields. That means the researchers were unable to map floods in smaller ones or in most cities. Clouds also interfered with the image processing algorithm, and since satellites passed over a specific location on Earth only once or twice a day, short-term floods were also missed.
Newer instruments have a much higher resolution and can see through clouds, he says. Bessie black, co-founder and CEO of Cloud to Street. These tools, along with artificial intelligence, can map floods with greater precision today. But to systematically map floods over time, the researchers had to stick to images from one source, using technology that has been around for longer.
The effort gives scientists a clearer picture than any other resource of the scale and human impact of the recent floods. And the results will be especially helpful for modelers trying to predict risk, he says. Philip Ward, who studies flood risk assessment at the Vrije Universiteit Amsterdam and was not involved in the study.